Revise bodyToStringParser to use MemoryDataStream (#2881)

Better dynamic memory reallocation and avoids conversion to MemoryDataStream at end anyway.
Still has the weakness that the decode is unbounded so bad actors can kill system with large POST.

Sming/Components/Network/src/Network/Http/HttpBodyParser.cpp

@@ -16,6 +16,7 @@
 
 #include "HttpBodyParser.h"
 #include <Data/WebHelpers/escape.h>
+#include <Data/Stream/MemoryDataStream.h>
 
 /*
  * Content is received in chunks which we need to reassemble into name=value pairs.
@@ -103,11 +104,11 @@ size_t formUrlParser(HttpRequest& request, const char* at, int length)
 
 size_t bodyToStringParser(HttpRequest& request, const char* at, int length)
 {
-	auto data = static_cast<String*>(request.args);
+	auto data = static_cast<ReadWriteStream*>(request.args);
 
 	if(length == PARSE_DATASTART) {
 		delete data;
-		data = new String();
+		data = new MemoryDataStream();
 		request.args = data;
 		return 0;
 	}
@@ -118,15 +119,10 @@ size_t bodyToStringParser(HttpRequest& request, const char* at, int length)
 	}
 
 	if(length == PARSE_DATAEND || length < 0) {
-		request.setBody(std::move(*data));
-		delete data;
+		request.setBody(data);
 		request.args = nullptr;
 		return 0;
 	}
 
-	if(!data->concat(at, length)) {
-		return 0;
-	}
-
-	return length;
+	return data->write(at, length);
 }